Looking at AI Adoption As More Than a Technology Challenge

AI adoption is not just a technology challenge. Psychology may be the more significant obstacle, writes Kevin Cohn of Brightflag. For legal ops teams accustomed to predictable software, AI’s variability, mistakes, and new ways of working require a fundamental adjustment in their expectations and behavior.

Traditional software has trained users to expect consistent results. Identical inputs should produce identical outputs while errors are treated as defects to be corrected. AI challenges these assumptions because the same prompt can generate different responses, and even sophisticated systems can produce obvious mistakes. 

That tension, however, does not mean AI is inherently untrustworthy. Cohn maintains that trust should instead be calibrated according to a system’s strengths, weaknesses, and performance on particular tasks. AI should be approached more like a person than conventional software. 

The psychological adjustment becomes even more significant when AI agents work simultaneously. Unlike humans, who lose efficiency when switching between demanding cognitive tasks, multiple AI agents can work on separate assignments at the same time. Yet overseeing those agents can create a management problem, requiring frequent input, approvals, and adaptation to different capabilities.

For legal ops teams, Cohn recommends a path forward that begins with experimentation. Teams should first push beyond their comfort zones to understand both AI’s capabilities and their own limitations, then make AI a habitual starting point for appropriate tasks. Only after that foundation is established should teams focus heavily on cost and quality optimization. 

As AI adoption is not just a technology challenge, Cohn emphasizes patience with both human and AI mistakes, arguing that experimentation, failure, and adjustment are essential to learning how to work effectively with AI.